Exchanging Information in Newsgroups

The economics of information are particularly important on the web where reproduction and distribution costs are remarkably small. This guest post describes new research on information distribution in newsgroups. The post was generously provided by one of the study's authors.

Identifying Discussion Leaders and Information Sources
by Itai Himmelboim¹

Here’s some old news: one of the greatest promises of the Internet is the ability for anyone with a PC and Internet connection to join forums on any topic imaginable and contribute, consume and exchange information and opinions. Information is available via a wide range of old sources (news websites) and new sources (blogs, forums, personal websites and news aggregators such as Google News).

Robert Nye said once that a richness of information leads to a poverty of attention. In a study I conducted with Marc Smith and Eric Gleave from Microsoft Research, the Netscan dataset was used to follow patterns of replies – indicators of attention – in 20 political newsgroups between July and December 2006.

It wasn’t a surprise to find that in all newsgroups, relatively few participants attracted a relatively large portion of the discussion to threads they started. After all, literature illustrates that large networks – be they of people, websites or even genes – tend to show a power-law distribution in which few participants receive a large and disproportional number of links – in our case, replies – from other participants. With that in mind, we began to explore the role that this small number of highly connected participants play in their groups.

We identified these highly replied participants using a range of statistical measures including: success in starting new threads, the percentage of all messages in a group that appeared in threads they started, as well as the percentage of individuals in the group that participated in these threads. We found only a handful of such highly connected participants in each group, making them less than one percent of the population in their newsgroups. Many of these participants attracted more than one-half of the discussion to threads they started. We decided to name them Discussion Catalysts, or DCs.

Deciding what to talk about

Discussion catalysts may not tell fellow group members what to say, but according to their attention grabbing records, they do tell groups what to talk about. Our next step was to determine what information they brought to the table.

Content analysis of messages that discussion catalysts used to start threads revealed an interesting phenomenon. If you thought, like I did, that political discussions in newsgroups start with an individual’s opinion, you may be surprised to find that this study shows otherwise. DCs play the role of content importers. They go outside their groups to the World Wide Web – news sites, blogs and other websites – search for interesting articles, and bring them to groups to discuss.

Another interesting finding was that although DCs import content from a range of sources, most of the articles came from traditional news sites such as the Washington Post and Associated Press. Less than one tenth of the entries came from blogs.

So what does all this tell us? First, although the Internet is free and open by its nature, when we interact freely, we tend to create a structure in which few of us get a lot and most of us get very little. Second, even when we use relatively new platforms for political discussions, the information comes from good (?), old news organizations. Why? Well, I’ll leave that for you to discuss.

¹ Itai Himmelboim is a doctoral student at the University of Minnesota's School of Journalism and Mass Communication.